01
Define the unit of work
Prefer a specific request or document to a vague objective such as automating an entire department.
How to choose your first AI project
Your first AI project needs clear boundaries.
A practical guide to choosing an AI use case with available data, a specific user and acceptance criteria.
THE STARTING POINT
The questions worth asking.
Start by describing a task as it is performed today: who does it, which information they use and what they deliver. Record exceptions and points where human judgement is essential. This helps select a small but representative first step.
FROM KNOWLEDGE TO PRACTICE
Better questions. Better-informed technology decisions.
We bring together starting points for the conversations that come before a project. Explore AI, cloud migration and security, and connect what you learn with your organisation's next steps.
Explore the guidesIN PRACTICE
Three ways to get started.
02
Check the data
Confirm which sources are available, who has access and whether examples of correct results exist.
03
Agree on evaluation
Measure quality, time and cost in representative scenarios, including failure cases.
USEFUL ANSWERS
Frequently asked questions
Let's answer the first questions.
Should the first project automate the entire task?
No. A solution that prepares a result for human review may be a better starting point.
When do we move beyond the pilot?
When the agreed criteria are met and support, access, monitoring and exception handling have been defined.
LET’S MOVE FORWARD.
Assess your first AI idea
Describe your environment and objective. We will start with what matters to your project.